跳到主要內容

臺灣博碩士論文加值系統

(44.211.31.134) 您好!臺灣時間:2024/07/24 16:16
字體大小: 字級放大   字級縮小   預設字形  
回查詢結果 :::

詳目顯示

: 
twitterline
研究生:鄭宇倫
研究生(外文):CHENG, YU-LUN
論文名稱:單光子斷層影像分類方法-帕金森氏症輔助診斷之應用
論文名稱(外文):Classification of SPECT Images - An Application for Assisting in the Diagnosis of Parkinson’s Disease
指導教授:朱基祥
指導教授(外文):CHU, CHI-HSIANG
口試委員:羅夢娜袁子倫
口試委員(外文):LO HUANG, MONG-NAYUAN, TZU-LUN
口試日期:2023-07-14
學位類別:碩士
校院名稱:東海大學
系所名稱:統計學系
學門:數學及統計學門
學類:統計學類
論文種類:學術論文
論文出版年:2023
畢業學年度:111
語文別:中文
論文頁數:57
中文關鍵詞:擴散映射流型距離拔靴法機器學習帕金森氏症單光子電腦斷層
外文關鍵詞:Diffusion Mapsdiffusion distancemachine learningParkinson’s diseaseSPECT
相關次數:
  • 被引用被引用:0
  • 點閱點閱:75
  • 評分評分:
  • 下載下載:11
  • 收藏至我的研究室書目清單書目收藏:0
帕金森氏症(Parkinson’s disease, PD)是一種常見的腦部退化疾病之一,在台灣帕金森氏病的盛行率,大約是每十萬人之中約有150至300人,盛行年齡以60歲至80歲為主,且隨著年紀越大,盛行率越高的一種神經退化性疾病。單光子電腦斷層(Dopamine Transporter Single Photon Emission Computed Tomography, DAT-SPECT)掃描顯影是主要用來診斷帕金森氏症的方法,傳統上仰賴醫師以目視方式進行初步影像判讀為主,但此方法易受醫師主觀經驗影響,特別是在症狀早期的時候不易準確判讀。而近年來機器學習應用於醫療影像上的技術日漸成熟,可以在許多領域上見到使用圖像辨識來進行輔助判讀。因此本研究著重在影像分類問題上,採用醫師從受試者DICOM影像挑選出的3張影像來進行研究分析。經由擴散映射(Diffusion Maps,DM)得到影像間的流型距離(Diffusion distance),利用傳統統計上常用的羅吉斯迴歸,以及一些常見的機器學習方法,如:線性判別分析、支持向量機、隨機森林、單純貝氏分類器及K-近鄰演算法等,來進行影像分類。
在本論文中,我們採用拔靴法(Bootstrapping)抽出100組樣本,建構出合適的分類器,並經由投票的方式增加模型整體的分類效果。同時分析經電腦定量後的12個圖像特徵,最後結合圖像與特徵分類的結果,做為輔助判讀的參考依據。

Parkinson’s disease (PD) is one of the common neurodegenerative disorders. In Taiwan, the prevalence of Parkinson’s disease is estimated to be approximately 150 to 300 individuals per 100,000 people. The prevalence is highest among individuals aged 60 to 80 years, and it increases with advancing age. Single-Photon Emission Computed Tomography (SPECT) imaging is a primary method used for diagnosing Parkinson’s disease. Traditionally, it relies on physicians visually interpreting the images, which can be subjective and particularly challenging for accurate interpretation during the early stages of the disease.
In recent years, the application of machine learning in medical imaging has become increasingly mature, and the use of image recognition for assisting in interpretation can be observed in various fields. Therefore, this study focuses on the problem of image classification, utilizing three selected DICOM images from subject as the basis for research analysis. The diffusion maps (DM) method is employed to obtain the diffusion distance between images. Traditional statistical methods such as logistic regression model, as well as commonly used machine learning algorithms including linear discriminant analysis, support vector machines, random forests, naive Bayes classifier, and k-nearest neighbors classifier, are utilized for image classification.
In this thesis, we employed the bootstrapping method to generate 100 sets of samples and constructed suitable classifiers. The overall classification performance of the models was enhanced through a voting mechanism. Additionally, we analyzed 12 quantified image features using computer-based techniques. Finally, we combined the results of image classification and feature classification for assisting in the diagnosis.

目錄 i
圖次 iii
表次 iv
中文摘要 v
Abstract vi
第一章、 緒論 1
1.1 前言 1
1.2 研究動機與目的 1
第二章、 研究資料簡介 3
2.1 DAT-SPECT影像檢視 3
2.2 研究資料介紹 5
第三章、 研究方法介紹 9
3.1 影像處理方法 9
3.1.1 影像預處理 9
3.1.2 擴散映射(Diffusion Maps, DM) 10
3.2 分類器介紹 15
3.2.1 羅吉斯迴歸 15
3.2.2 線性判別分析 15
3.2.3 支持向量機 17
3.2.4 隨機森林 20
3.2.5 單純貝氏分類器 21
3.2.6 K-近鄰演算法 22
第四章、 資料研究流程與結果分析 23
4.1 研究流程 23
4.2.1 630影像單一模型比較 26
4.2.2 加入輔助標籤方法之混淆矩陣 29
第五章、 包含定量特徵資料研究流程與結果分析 33
5.1 12個特徵判別模型 33
5.1.1 12特徵判別模型比較 37
5.1.2 413影像單一模型比較 39
5.1.3 加入輔助標籤方法之混淆矩陣 43
5.2 影像模型與特徵模型綜合分析 45
第六章、 結論與未來展望 50
參考文獻 51
附錄 53

[1] Arena, J. E., Urrutia, L., Falasco, G., Leon, M. P. D., Vazquez, S., Rossi, M., & Merello, M. (2021). Correlation between 99m Tc-TRODAT-1 SPECT and 18 F-FDOPA PET in patients with Parkinson’s disease: a pilot study. Radiologia Brasileira, 54, 232-237.
[2] Bah, B. (2008). Diffusion maps: analysis and applications.
[3] Ding, J. E., Chu, C. H., Huang, M. N. L., & Hsu, C. C. (2021). Dopamine Transporter SPECT Image Classification for Neurodegenerative Parkinsonism via Diffusion Maps and Machine Learning Classifiers. In Medical Image Understanding and Analysis: 25th Annual Conference, MIUA 2021, Oxford, United Kingdom, July 12–14, 2021, Proceedings 25, pp. 377-393. Springer International Publishing.
[4] De la Porte, J., Herbst, B. M., Hereman, W., & Van Der Walt, S. J. (2008). An introduction to diffusion maps. In Proceedings of the 19th symposium of the pattern recognition association of South Africa (PRASA 2008), Cape Town, South Africa, pp. 15-25.
[5] Jankovic, J. (2008). Parkinson’s disease: clinical features and diagnosis. Journal of neurology, neurosurgery & psychiatry, 79(4), 368-376.
[6] Nadler, B., Lafon, S., Kevrekidis, I., & Coifman, R. (2005). Diffusion maps, spectral clustering and eigenfunctions of Fokker-Planck operators. Advances in neural information processing systems, 18.
[7] Pedregosa, F., Varoquaux, G., Gramfort, A., Michel, V., Thirion, B., Grisel, O., ... & Duchesnay, E. (2011). Scikit-learn: Machine learning in Python. the Journal of machine Learning research, 12, 2825-2830.
[8] Rocca, W. A., Grossardt, B. R., & Maraganore, D. M. (2008). The long-term effects of oophorectomy on cognitive and motor aging are age dependent. Neurodegenerative diseases, 5(3-4), 257-260.
[9] Zachry, J. E., Nolan, S. O., Brady, L. J., Kelly, S. J., Siciliano, C. A., & Calipari, E. S. (2021). Sex differences in dopamine release regulation in the striatum. Neuropsychopharmacology, 46(3), 491-499.
[10] Zhou, Z. H. (2012). Ensemble methods: foundations and algorithms. CRC press.

QRCODE
 
 
 
 
 
                                                                                                                                                                                                                                                                                                                                                                                                               
第一頁 上一頁 下一頁 最後一頁 top
無相關期刊